Risk Assessment of Postoperative Nausea and Vomiting. Christian C. Apfel, MD Norbert Roewer, MD

Transcription

1 Risk Assessment of Postoperative Nausea and Vomiting Christian C. Apfel, MD Norbert Roewer, MD Introduction It is generally accepted that emetic sequelae such as postoperative nausea, postoperative vomiting, and postoperative nausea and vomiting (PONV) are of multifactorial origin and that the incidence of PONV after balanced anesthesia despite the advances in the last decades is still between 20% and 30%. 1,2,3 Anaesthetic factors such as volatile anesthetics and opioids, the type of surgery, and patient-related factors are all thought to have a major impact on PONV. 4,5 Thus, for a long time it was assumed that prediction of PONV is difficult, if not impossible, as PONV was believed to be influenced by too many factors. However, there is now sufficient evidence that the risk for emetic sequelae after balanced anesthesia in adults can be predicted by a risk score derived from a limited number of risk factors In order to reduce the incidence of PONV, antiemetic strategies such as choosing a less emetogenic anesthetic technique, 12 applying antiemetics, 13 or both by a multimodal approach 14 can be applied. Because nonselective routine prophylactic antiemetic treatment may not improve patient outcome when applied to all patients, 15 the use of a risk-adapted prophylactic strategy would be the most rational approach. 16 These issues have been discussed in a review by Everett in the Spring 2002 issue of this journal. 17 However, very recently there have been some additional insights and advances that will be updated in this chapter. Thus, this review provides an update on risk factors for PONV. It will explain how risk models based on true risk factors could be developed. Further, the applicability but also the limitations of predictive models will be discussed. In addition, a brief summary of antiemetic strategies needs to be discussed. Finally, a risk-dependent antiemetic strategy based on current evidence and easily implemented in daily clinical practice will be suggested. 13

2 14 Apfel and Roewer Which are the True Risk Factors? It is appreciated that certain conditions are associated with high incidences of PONV, e.g., a patient s history of PONV, females, gynecological surgery, and long operations. This led to the widely accepted but false belief that these conditions are risk factors. Because PONV is likely to be of multifactorial origin, the consideration of a high incidence of a single factor without correcting for other coexisting factors may be misleading. For example, the average incidence of PONV after various types of surgery is 30%, and females suffer 3 times more PONV than men. Consequently, the average incidence of PONV would be 45% and 15% for females and males, respectively. Most surgical specialties with female and male patients would have an incidence in the range of 30%. However, by just considering that gynecological surgery is usually performed in females, we would expect the average incidence of PONV to be in the range of 45% (almost every second patient!). Thus, the attempts to explain the higher incidences of this example after gynecological surgery do not require complex pathophysiological speculations such as irritation of the autonomous innervation of the female reproduction organs. Interestingly, such pathophysiological hypotheses are not applied for urologic surgery despite the presence of similar innervations, and the considerably lower incidence has never drawn much attention. This example demonstrates that the incidences on their own must be evaluated with the full realization that several extraneous factors may be exerting their influence at the same time, as already pointed out almost 50 years ago by Dent et al. 18 A high incidence associated with a certain condition, such as type of surgery, is therefore no proof for a relevant risk factor. This has important clinical implications when risk-adapted prophylactic antiemetic strategy is to be established. Because risk assessment should be based on the relative impact of true risk factors, i.e., independent risk factors with clinical importance, the following risk factors are classified accordingly. Risk Factors With a Sufficient Body of Evidence Patient-Specific Risk Factors Female gender has been associated with a higher incidence compared to male patients, 19,20 but more importantly, also proved to be an independent predictor in multivariable analyses On average, females patients suffer three times more often from PONV than men, and it is therefore the strongest patient specific predictors. Cohen et al. were the first who identified nonsmokers to suffer more frequently from PONV than smokers. 21 These observations have been confirmed by most other studies. 8,11,22 24

3 Risk Assessment of PONV 15 The third patient-specific risk factor is the individual disposition of the patient as expressed by the patient s history of motion sickness or PONV. 6 11,19 Some use this risk factor alone as a guide to estimate the patients risk for PONV. However, the overall predictive value is not as sensitive as using a simplified risk score. 11 Age has a nonlinear impact on PONV. Although postoperative vomiting occurs infrequently in neonates and toddlers, the incidence increases to 20% to 40% in school children with little impact on the gender before puberty The incidence is highest in young adults and decreases continuously in the elderly. 8,10,21 However, the impact is not as strong as the previously mentioned risk factor, which explains why it failed statistical significance in some multivariable analyses. 7,11 Anesthetic Techniques and Drugs There is a general agreement that PONV is rare after uncomplicated peripheral nerve blocks, although studies with PONV as their primary outcome are lacking. Some studies also suggest a lower incidence after regional anesthesia compared to general anesthesia. 10,28 In a study for breast surgery, a paravertebral nerve block technique was associated with less PONV compared to a propofol-based general anesthesia. 29 Generally, orthopedic surgery is assumed to be associated with a lower risk for PONV because of less emetogenic autonomic stimulation compared to abdominal surgery. However, another explanation may be that regional anesthetic techniques are more widespread in orthopedic surgery, leading to a lower overall incidence of PONV. It should be noted that in a study with a majority of orthopedic patients undergoing general anesthesia, the actual incidences of PONV correlated well with those predicted by an operation independent risk score. 30 And, even more, in a multivariable analysis of a prospective survey in ambulatory, orthopedic surgery and especially shoulder surgery had a higher risk for PONV. 10 The use of a balanced anesthetic technique is known to be associated with about twice as much PONV compared to the use of a total intravenous anesthesia. 31 Because antiemetic effects of subhypnotic doses of propofol were demonstrated in patients suffering from chemotherapyinduced nausea and vomiting, 32 and this was corroborated by studies regarding the treatment of established PONV, 33 propofol is generally assumed to be antiemetic. 34 However, the antiemetic effect was apparently short lived. 32,33 Consequently, Scuderi et al. investigated a rather small bolus of 0.1 mg/kg followed by a continuous infusion with 1 mg/kg/h propofol but with no antiemetic effect. 35 Gan et al. developed the concept of a patient-controlled antiemesis where patients who had symptoms of nausea could press a button to receive 20 mg propofol. 36 Based on these data, the simulated ED50 plasma concentration of propofol was 343 µg/ml for relieving nausea in 50% of patients.

4 16 Apfel and Roewer In a recent randomized controlled trial of factorial design with 1180 patients, volatile anesthetics were associated with twice as much PONV compared to propofol anesthesia. 12 This result was very similar to the majority of previous studies as corroborated by the meta-analysis from Sneyd et al. 31 However, a detailed analysis of the time course to PONV by Kaplan-Meier charts revealed that the main difference occurred in the first 2 hours (Fig. 1). Further analysis illucidated that the incidence of early PONV correlated strongly with the degree of exposure to volatile anesthetics (dose response relationship), whereas no effect could be demonstrated for propofol anesthesia (Fig. 2). 12 (ie, are true and independent predictors.) This is highly suggestive that volatile anesthetics not only correlated with but caused PONV. In patients at very high risk of PONV, it may therefore make better sense to avoid volatile anesthetics in the first place. A majority of studies showed a higher incidence of PONV when nitrous oxide was used. 37,38 However, it is more appropriate to quantify the emetogenic effect by a meta-analysis, which resulted in a relative risk of 1.4 for nitrous oxide. 39 Because this effect appears much less than the impact of the previously mentioned volatile anesthetics, the clinical importance for PONV may have been overestimated. Opioids are well accepted to cause nausea and vomiting. However, the mechanisms of opioids are complex. In ferrets, administration of 0.1 Figure 1. Kaplan-Meier curves representing the proportion of patients who vomited (n = 941). In order to ensure the comparability with propofol, patients receiving volatile anaesthetics but no intraoperative opioids (n = 239) were excluded. Note that the difference between propofol and volatile anaesthetics is related only to the early postoperative period. 11

5 Risk Assessment of PONV 17 Figure 2. Correlation between the degree of exposure to anesthesia and early postoperative vomiting. In order to be able to compare inhalational and propofol anesthesia, five percentile groups were formed for each as a function of the anesthesia duration. Note that the incidence of early vomiting correlates positively with the degree of exposure to inhalational anaesthesia but not to propofol anaesthesia. 11 mg/kg of morphine triggers emesis, and an increased dose up to 0.2 mg/kg triggers emesis in all animals. It is believed that activation of the chemoreceptor-trigger zone (CTZ; located in the area postrema) stimulates the vomiting center (VC) in the brain stem (i.e., formatio reticularis, tractus nucleus solitarius) because ablation of the CTZ prevents morphine-induced vomiting. However, larger doses reduce vomiting, and 5 mg/kg morphine can completely block emesis. 40 Apomorphine is a pure dopamine agonist with no activity on opioid receptors and is known to induce emesis by stimulating the CTZ. Interestingly, the emetic effect of apomorphine can be blocked dose dependently by the administration of morphine, fentanyl, and methadone, and these opioid induced antiemetic effects in the VC can be antagonized by naloxone. 41 Thus, low doses of opioids are emetogenic by stimulating the dopamine receptors in the CTZ but high doses may exert an antiemetic effect by stimulation of µ-receptors in the VC. However, such a bell-shaped dose response has never been unequivocally proven in humans, and studies suggest that an increase of opioids increases the risk for vomiting. 42,43 In addition, comparisons between the opioids, e.g., fentanyl and alfentanil, 44 fentanyl and

6 18 Apfel and Roewer sufentanil, 45 morphine, pethididine, fentanyl, and sufentanil, 46 resulted in contrary results so that there may not exist clinically significant differences between the studied opioids. 47 In a recently published paper alfentanil, when compared to fentanyl or sufentanil led to a lower incidence of PONV. 48 However, low-dose opioids for induction of anesthesia (1.5 µg/kg fentanyl, 15 µg/kg alfentanil, or 0.15 µg/kg sufentanil) led only to a small and not significant increased incidence of PONV. 12 Therefore, further research in this field is warranted to clarify this issue. While data on the impact of opioids given intraoperatively are controversial, several studies suggest that the use of postoperative opioids increases the risk for PONV. 7,11,12,21,49 Of note, there seems to be no difference in the incidence of PONV when morphine or piritramide is used for patient controlled analgesia. 13 Risk Factors With Conflicting Evidence Type and Site of Surgery Early studies indicated that the incidence of PONV varies with the type and site of surgery, 18,20,50,51 and this seems to be still true for various types of surgery, e.g., strabismus surgery, gynecological surgery, cholecystectomy, and intracranial surgery. 5,52 However, recent large prospective trials in adults using multivariable analyses, as outlined above, came to conflicting results. Cohen et al. found that gynecological laparoscopy and ophthalmologic surgery were independent predictors (i.e., risk factors) for postoperative nausea, whereas abdominal surgery and ENT surgery failed statistical significance. 21 It is also interesting to note that in the analysis of female patients, gynecological surgery (intubated, nonlaparoscopic) was not an independent risk factor for nausea. This is corroborated by a Finnish survey from Koivuranta et al. 7 Although the highest incidence was observed in gynecological patients (52%), the type of surgery was not an independent risk factor for PONV in that population. 7 At the same time, data collected from ENT patients also suggested that type of surgery may not be an independent risk factor for PONV. 8 When the same study was performed in general surgery and ophthalmologic patients, the type of surgery turned out to be a significant risk factor when the duration of anesthesia was not considered. 9 When the data of all three studies were taken together, again the type of surgery was no longer an independent risk factor. 11 However, a large survey in outpatients suggested that several types of surgery especially orthopedic surgery were independent predictors of PONV. 10 In a recent analysis of predictors of the need for rescue antiemetic in the postanesthesia care unit (PACU), the only surgical association was the duration of surgery (which correlates strongly with the duration of anesthesia), but none of the types of surgery turned out to be

7 Risk Assessment of PONV 19 Table 1. Risk Factors for PONV Positive results with strong clinical impact Mostly positive results Conflicting results Flawed or insufficient data Disproved factors Female gender Nonsmoking status History of motion sickness or PONV General versus regional anesthesia Volatile anesthetics Duration of anesthesia Postoperative opioids Young age and physical status Nitrous oxide is emetogenic whereas 80% oxygen seems to be antiemetic Muscle relaxant antagonists Site of surgery Menstrual cycle Experience of the anaesthetist Gastric tube for decompression Pain Movements Body mass index Anxiety and personality The classifications are made on the strengths of evidence and clinical impact ( = modest; = strong; = very strong). PONV = postoperative nausea and vomiting. statistically significant. 53 However, it should be kept in mind that this study was not prospectively designed for PONV and some methodological concerns (e.g., patient s history of PONV was not considered) were present. Due to the present controversy, the type and site of surgery does not Table 2. Risk Table for Postoperative Vomiting According to the Patient s Gender, Smoking Status, History of PONV or Motion Sickness, and Age (Under the Condition of a 2-Hour Operation) 7 Gender Smoking Status History of PONV or Motion Sickness Age in Yrs Male Yes Negative Male Yes Positive Male No Negative Male No Positive Female Yes Negative Female Yes Positive Female No Negative Female No Positive PONV = postoperative nausea and vomiting.

8 20 Apfel and Roewer appear to provide consistent and reliable additional information to assess the patients risk for PONV. Although the above mentioned applies to adults undergoing balanced anesthesia, this may not be true for children. For example, strabismus surgery in adults has been reported to be associated with a surprisingly low incidence of PONV, 54 whereas in children, the incidence of postoperative vomiting is very high. 5,55 In this context, Rusch et al. were the first to describe that the posterior fixation suture ( faden-operation ) technique was associated with a higher incidence of PONV compared to the more commonly used recessions technique. 56 Two years later, the same observation was described by another study group 57 so that it is reasonable to assume that at least for strabismus surgery in children the surgical technique may contribute to postoperative vomiting. However, there is a lack of evidence for other surgical types or techniques and as outlined above a simple comparison of the raw incidences may be flawed by confounding factors so that multivariate analyses are also needed. But unfortunately, the surveys in pediatric patients are restricted to the simple reporting of incidences so that no firm conclusions regarding the impact of the type of surgery can be drawn. 25,26 Therefore, whether the type of surgery is a risk factor of postoperative vomiting in children is not clear but seems likely. 27 In addition, according to a very recent analysis using a Dale Model, it became clear that risk factors for nausea and vomiting may not be the same. 14 While the patient specific risk factors, e.g. female gender, seem to influence both symptoms, the type of surgery was not a risk factor for vomiting but for nausea. Whether this occurs due to circular reasoning while nausea is assessed because PONV is expected to be more frequent after certain operations, or whether this has really pathophysiological reasons, needs to be further investigated. Mask Ventilation, Experience of the Anesthesiologist, Gastric Tube Around 1960, two studies described higher incidences of PONV after ventilation with a face mask compared to intubation, whereas an earlier study was unable to find such an effect. 18,20,51 Because face mask ventilation can cause gastric distension and the latter can elicit the vomiting reflex, 58 it was easily accepted that face mask ventilation causes PONV. 1 In addition, a Finnish group, Hovorka et al., described a higher incidence of PONV when face mask ventilation was performed by a nurse anesthetist. 59 However, Hechler et al. randomized 669 patients to be ventilated by a face mask before intubation (3 min) or not. 60 The incidence of postoperative vomiting and PONV was similar between the groups and analyzing the experience of the anesthesiologist did not show any impact on PONV. 60

9 Risk Assessment of PONV 21 As the Finnish group assumed that gastric inflation may be a main trigger of PONV, they randomly assigned patients in their following study to decompression of the gastric contents via a nasogastric tube or without. But the incidence of PONV was similar. 61 Interestingly, one study showed even an increased incidence of PONV when a nasogastric tube was applied. 62 In a meta-analysis of about 4000 patients, a nasogastric tube was associated with a slightly lower incidence of postoperative vomiting but the incidence of postoperative nausea remained the same (not to mention that a nasogastric tube was associated with a two-fold incidence of atelectasis, fever, and pneumonia). 63 These examples demonstrate that a potential influence is easily assumed to be a risk factor when a mechanism appears plausible. So, if trials demonstrated that a nasogastric tube has no impact on postoperative nausea (e.g., for average adult patients receiving standard care), our judgment should be guided more by the existing data than by plausible hypotheses. At the same time, this may not be true in special circumstances (e.g., mask ventilation in children), however, there is no evidence for that. Body Mass Index Numerous mechanisms have been claimed to cause PONV in overweight patients (difficult mask ventilation, increased storage etc.). 55 However, Lerman concluded that the current literature provides insufficient evidence for such a relationship. 5 Since then, numerous multivariate analysis of large surveys were performed. Only one survey identified body mass index >25 kg/m 2 to be associated with an approximately 1.5-fold risk, 7 whereas all other analyses were unable to identify the body mass index as an independent predictor for PONV. 8 12,21,48,64 In addition, a systematic review revealed that besides errors in citations, there is good evidence to conclude that the body mass index has no impact on PONV. 65 Models (Scores) for Clinical Risk Assessment The Development of Risk Scores The solution of this multivariable problem is the application of a special statistical method called multiple logistic regression analysis. Without going too much into detail, it is important to know that this method allows determination of the relative impact of all factors included into the model at the same time. Such a method was first applied by Palazzo and Evans, who identified female gender, the patient s history of PONV, history of motion sickness, and the use of postoperative opioids. 6 Unfortunately, the study was restricted to minor orthopedic surgery so that a potential relative impact of the operation could not be calculated.

10 22 Apfel and Roewer A few months later, a meta-analysis of factors affecting the incidence of nausea and vomiting after gynecological surgery was published in the same journal. 66 This study from Haigh et al. (Glaxo Ltd., UK) revealed that the most important factors for PONV were the antiemetics (ondansetron 1, 8, or 16 mg), the type of operation, and the neuromuscular blocking agents. Unfortunately, the analysis did not consider previously established risk factors such as female gender (because only females were analyzed) or the history of motion sickness or previous PONV so that the results must be interpreted with caution. One year later, Cohen et al. published a study focussing on a postoperative interview in four centers with approximately 16,000 patients. 21 The main risk factors were female gender, young age, a good physical condition as reflected by the American Society of Anesthesiologists (ASA) physical status, nonsmoking, and a long duration of anesthesia for postoperative nausea as well as for postoperative vomiting (separate analyses). They also identified marked variations in the incidence of emetic sequelae across hospitals, which was assigned as a center effect. However, this may be caused by a more or less frequent use of prophylactic antiemetics, which most regrettably were not recorded in the study. In 1996, the risk table developed by Palazzo and Evans was applied to 400 patients with different operations by a simple calibration curve of predicted and actual incidences of PONV and creation of 16 risk groups. A good correlation between the previously developed risk table and the incidence of PONV could be demonstrated, but a correct prediction in individuals was only 71% and therefore not perfect. 67 Unfortunately, a multivariable recalculation was not performed. A multivariable analysis has been done by Koivuranta et al., who identified female gender, nonsmoking, a history of PONV, a history of motion sickness, and a long duration of anaesthesia as risk factors for PONV. 7 Based on the coefficients of the significant variables, a new risk score was developed. The novelty of their analysis was the simplification of the risk score. Instead of calculating the precise risk by an exponential equation, the simplified score consisted of the number of the five most important risk factors. These were female gender, nonsmoking status, history of motion sickness, history of PONV, and a duration of surgery >60 min. Simultaneously, Apfel et al. 68,69 performed a large prospective survey in 1137 elective patients undergoing otolaryngological procedures. 8 Again, patient-related factors and the duration of anesthesia were most important for the prediction of postoperative vomiting. As the type of operation had little impact on postoperative vomiting, an operation-independent risk could be calculated using the coefficients of the logistic equation: risk probability of PV = e z,

11 Risk Assessment of PONV 23 where z = 1.28*(female gender) 0.029*(age) 0.74*(smoking) *(history of motion sickness or PONV) *(duration) 0.92 when the variables are coded as follows: gender (male = 0, female = 1), age (in decades), smoking status (no = 0, yes = 1), history of motion sickness and/or PONV (no = 0, yes = 1), and duration of anesthesia (in hours, e.g., 45 min = 0.75 h). As this study was performed in ear, nose, and throat (ENT) patients, the question arose whether such a score is also applicable to patients undergoing general surgery (Score I). Subsequently, over 1,000 patients undergoing general surgery, orthopaedic surgery, vascular surgery, and squint surgery were prospectively studied. The multivariable analysis revealed that at the expense of the duration of anesthesia some types of surgery were statistically significant predictors of postoperative vomiting. Thyroid surgery was associated with an increased risk of 2.7 ( ), whereas laparoscopic surgery was associated with a decreased risk of 0.36 ( ). Therefore, a new score was developed (which considered the type of surgery) and was compared with the previous operation-independent risk score (Score II). Interestingly, the prediction of both scores was clinically very similar so that for the sake of a more general applicability the operation-independent risk score was preferred (Score I). To ease the clinical applicability, a risk table for male and female patients has been published so that no complicated calculation is required for daily practice. 8 To evaluate the applicability of risk scores between different countries, two centers (Würzburg, Germany and Oulu, Finland) decided to reanalyze their data and to perform a cross-validation to answer the following questions: to what extent are the risk scores specific for their centers, and are they applicable to a different center? How strong is a possible center effect in a combined data set? Can regression-based risk scores be simplified without loss of accuracy? 11 Several analyses were performed to set up risk scores for each center and for a combined data set. All developed risk scores could estimate the patient s risk for PONV in each center. Although the raw incidences and population of the centers appeared to be very different, the center itself did not affect PONV if corrected for the relevant predictors (no center effect). Further, a simplification of the scores based on equally weighted risk factors did not affect its accuracy so that the final score considered four equally weighted predictors: female gender, history of motion sickness and/or PONV, nonsmoking, and the use of postoperative opioids. In adults undergoing balanced anesthesia, the risk for PONV was 10, 20, 40, 60, or 80% when 0, 1, 2, 3, or 4 of the four factors are present, respectively (Fig. 3). 11 At the same time, Sinclair et al. analyzed data recorded from over 17,000 ambulatory patients. 10 Applying a similar analysis as described previously, 8 they developed a similar formula to calculate the risk for PONV.

12 24 Apfel and Roewer Figure 3. Correlation of the average predicted and actual incidence of PONV when the risk score based on both populations is applied. Note that the different incidence of PONV between the centers were due to the unequal distribution of risk factors, so that the same score was well applicable to both centers (data from 10 ). It is interesting to note that the type of surgery (orthopedic, otolaryngeal, and gynecological surgery) had a significant impact on PONV. One advantage of that score is that a much wider spectrum was studied. It should be noted that general anesthesia was associated with about a 10-fold higher incidence compared to local/regional anesthesia, an effect which is several times stronger than the other identified risk factors. Meanwhile, several other risk scores have been suggested. However, most of them have methodological difficulties that make a general applicability to other centers questionable, so they are not discussed here. 49,70 Validation of Risk Scores Most risk scores except the simplified risk score from Apfel et al. 11 were developed in a single center under specific conditions. This calls for external validations in other centers. Toner et al. 67 were the first who validated the risk model developed by Palazzo et al. 6 in 400 patients undergoing various types of surgery. A good correlation was found between predicted risk classes and the actual incidence, but the overall predictive value was only 71% (this means that the prediction, whether a patient will suffer PONV or not, was correct in 71%). Other validations came to similar results, although they used the more appropriate method of calculating the area under the receiver operating characteristic (ROC) curve. 9,30,71 A ROC curve plots all corresponding pairs of sensitivity and specificity for the whole range of possible risks (where a patient would be

13 Risk Assessment of PONV 25 classified to suffer PONV). The area under the ROC curve (AUC) is a global measure of the validity of a risk score for that specific population. It is usually between 0.5 (random prediction) and 1.0 (perfect prediction). Most scores are in the range of 0.65 to 0.7, which means in practice that approximately 70% of patients are correctly classified. This was somewhat disappointing, and a computer simulation found that due to the limited strength of the single predictors (the relative risk for female gender is approximately 3 and for the other risk factors in the range of 2), an overall correct prediction will at best be in the range of 70% to 80% (unless a much stronger predictor is found). 72 These results were finally confirmed by two independent studies. 49,64 It is interesting to note that the two simplified risk scores from Koivuranta et al. 7 and Apfel et al., 11 which did not consider the type of surgery, offered the best clinical prediction. However, it should be noted that all validations were performed in inpatients. Thus, it is possible that the score from Sinclair et al. provides better predictions in outpatients on which the original calculations were made. 10 In conclusion, although the patient s risk can be well estimated using risk scores, the prediction whether a certain individual will suffer from PONV will only be correct in approximately 70%. However, this is much higher than classifying patients according to their history of PONV or the type of surgery they are going to have. Because the simplified risk scores appeared to have the best predictive properties, 6,10 while still being easy to use (e.g., Fig. 4), this can be used for a risk dependent antiemetic strategy. Prophylactic Strategies to Reduce Postoperative Nausea and Vomiting There is increasing agreement among anesthesiologists that unselective use of antiemetics is of doubtful benefit. First, patients satisfaction does not appear to be superior when compared with immediate rescue treatment. 15 Secondly, antiemetics may cause unnecessary costs and side effects in those patients who do not suffer from PONV anyway. As suggested by Watcha, it is therefore sensible to use a risk-dependent antiemetic approach. 16 In patients at low risk, it is reasonable to wait and treat if it becomes necessary. If the patient s risk increases, it does make sense to lower the baseline risk and/or use one or a combination of antiemetics so that the incidence can be decreased to an acceptable low rate. Lowering the Baseline Risk Cisplatin is the strongest risk factor for chemotherapy-induced nausea and vomiting (CINV). This risk factor is several times stronger than the patient-specific risk factors such as female gender and a positive history of nausea and vomiting in previous chemotherapy cycles. In PONV, some

14 26 Apfel and Roewer Figure 4. A risk-dependent antiemetic approach. Depending on the personal preferences, other approaches may be reasonable. Compare also with the perhaps more cost-effective approach suggested by Watcha. 15 parallels can be found. For example, although there are only few randomized controlled trials, PONV is an extremely rare event after peripheral regional anesthesia. 28,29 This is also supported by the survey from Sinclair et al. where general anesthesia was associated with an approximately 10-fold risk compared to local/regional anesthesia. 10 Avoiding general anesthesia if suitable in that case may therefore be the best way to prevent PONV. If general anesthesia is needed, a total intravenous anesthesia (TIVA) with propofol is associated with a much lower incidence of PONV than a balanced anesthetic technique with volatile anesthetics For inpatients, an easy and apparently reliable way to estimate the individual risk for PONV is the use of a simplified risk score 11,49,64 because this allows a more appropriate classification than the single use of the patient s history of the type of surgery (unpublished observation). In addition, medical aspects (a patient with a wired jaw must not vomit) and personal preferences (patient wants to have as few drugs as possible) should be considered in the decision. 2. When the patient s risk is low (e.g., no risk factors, 10%), even if an antiemetic strategy would be highly effective, the numberneeded-to-treat (NNT) will be 10 (out of 100 patients, 90 will not

15 Risk Assessment of PONV 27 suffer PONV anyway and no more than 10 out of 100 will benefit, i.e., 1 out of at least 10 patients will benefit). 3. Prophylaxis is justified in patients with a moderately increased risk (e.g., risk score 2, 40%). There are different opinions on whether a single antiemetic or a combination should be used. From quantitative systematic reviews, it became apparent that the single use of an antiemetic decreases the event rate to approximately 0.7, i.e., the relative reduction rate (RRR) of a single antiemetic is in the range of 30%. With a combination of two antiemetics, the event rate can be halved (0.7 * 0.7 = 0.49). We prefer dexamethasone as the drug of first choice because it is cheap 90 and no side effects are known from the single use of up to 10 mg. 73 The combination with a 5-HT 3 antagonist, such as ondansetron, can further decrease the incidence. 74,75 In addition, the study from Splinter and Rhine suggests that in combination with dexamethasone, the dosage of ondansetron can be significantly decreased. 76 However, large dose response studies of antiemetic combinations are lacking. 4. When the risk of PONV is very high (e.g., risk score of 3 or 4, 60 or 80%), it seems reasonable to avoid volatile anesthetics and nitrous oxide and to perform a total intravenous anesthesia (TIVA) with propofol. Even if a TIVA is performed, the incidence of PONV will still be unacceptably high (perhaps due to the persistent need of intraoperative and postoperative opioids). Therefore, it would make sense to perform the TIVA with the abovementioned combination of dexamethasone and ondansetron. 5. If a patient suffers PONV despite being at low risk or despite prophylactic means, rescue antiemetics should be given, because the recurrence rate is usually >50%. Generally speaking, all prophylactic antiemetics can be given as rescue treatment, but dexamethasone may be an exception because of the slow (but long lasting) onset of antiemetic action. If a patient is at low risk and thus did not receive antiemetics, a 5HT 3 antagonist can be the first choice. One advantage of rescue versus prophylactic treatment is that much lower doses are needed, e.g., only 1 mg instead of 4 mg ondansetron. 77 This is in accordance with the observation that ondansetron is not effective in the therapy of established PONV when it has been given prophylactically before. 78 Although there are not further recent studies, it is generally assumed that if one antiemetic fails, an additional antiemetic acting via a different receptor or mechanism should be used. Conclusion Postoperative nausea and vomiting has a multifactorial etiology. Numerous risk factors have been described but only few seem to be unequivo-

16 28 Apfel and Roewer cally proven. Female gender, the patient s history of PONV or motion sickness, nonsmoking status, volatile anesthetics, nitrous oxide, and opioids have shown to be independent predictors for PONV in various centers. Several risk scores were developed to predict PONV. For adult inpatients undergoing balanced anesthesia, a simplified risk score based on the number of the four risk factors (female gender, history of PONV or motion sickness, nonsmoking status, and expected need for postoperative opioids) seem to provide a valid risk assessment. When 0, 1, 2, 3, or 4 of the four factors are present, the patient s risk for PONV is about 10, 20, 40, 60, or 80%, respectively. Prophylaxis may not be justified in patients at low risk. However, prophylaxis with dexamethasone and a serotonin antagonist can be a reasonable approach in patients at increased risk (e.g., risk score 2, 40%). In patients at high and very high risk (e.g., risk score 3 or 4, 60 to 80%), volatile anesthetics and nitrous oxide should be avoided by using a total intravenous anesthetic technique in conjunction with dexamethasone and a scrotonin antagonist. Rescue treatment should be given to every patient because the rate of recurrence is usually high (>50%). Doses required for the rescue treatment of established PONV may be half or even a quarter of that needed for prophylaxis. It is generally assumed that if one antiemetic fails, an antiemetic acting via another mechanism should be given. References 1. Palazzo MGA, Strunin L. Anesthesia and emesis. 1. etiology. Can Anaesth Soc J 1984;31: Kovac AL. Prevention and treatment of postoperative nausea and vomiting. Drugs 2000; 59: Watcha MF. Postoperative nausea and emesis. Anesthesiology Clin N Am 2002;20: Rabey PG, Smith G. Anaesthetic factors contributing to postoperative nausea and vomiting. Br J Anaesth 1992;69:40S 45S 5. Lerman J. Surgical and patient factors involved in postoperative nausea and vomiting. Br J Anaesth 1992;69:24S 32S 6. Palazzo M, Evans R. Logistic regression analysis of fixed patient factors for postoperative sickness: a model for risk assessment. Br J Anaesth 1993;70: Koivuranta M, Laara E, Snare L, et al. A survey of postoperative nausea and vomiting. Anaesthesia 1997;52: Apfel CC, Greim CA, Haubitz I, et al. A risk score to predict the probability of postoperative vomiting in adults. Acta Anaesthesiol Scand 1998;42: Apfel CC, Greim CA, Haubitz I, et al. The discriminating power of a risk score for postoperative vomiting in adults undergoing various types of surgery. Acta Anaesthesiol Scand 1998;42: Sinclair DR, Chung F, Mezei G. Can postoperative nausea and vomiting be predicted? Anesthesiology 1999;91:

17 Risk Assessment of PONV Apfel CC, Laara E, Koivuranta M, et al. A simplified risk score for predicting postoperative nausea and vomiting: conclusions from cross-validations between two centers. Anesthesiology 1999;91: Apfel CC, Kranke P, Katz MH, et al. Volatile anaesthetics may be the main cause for early but not delayed postoperative nausea and vomiting: a randomized controlled trial of factorial design. Br J Anaesth 2002;88: Tramer MR. A rational approach to the control of postoperative nausea and vomiting: evidence from systematic reviews. Part I. efficacy and harm of antiemetic interventions, and methodological issues. Acta Anaesthesiol Scand 2001;45: Scuderi PE, James RL, Harris L, et al. Multimodal antiemetic management prevents early postoperative vomiting after outpatient laparoscopy. Anesth Analg 2000;91: Scuderi PE, James RL, Harris L, et al. Antiemetic prophylaxis does not improve outcomes after outpatient surgery when compared to symptomatic treatment. Anesthesiology 1999;90: Watcha MF. The cost-effective management of postoperative nausea and vomiting [editorial; comment]. Anesthesiology 2000;92: Everett LL. Can the risk of postoperative nausea and vomiting be identified and lowered during the preoperative assessment? Int Anesthesiol Clin 2002;40: Dent SJ, Ramachandra V, Stephen CF. Postoperative vomiting: incidence, analysis and therapeutic measures in 3000 patients. Anesthesiology 1955;16: Burtles R, Peckett BW. Postoperative vomiting some factors affecting its incidence. Br Med J 1957;29: Bellville JW, Bross IDJ, Howland WS. Postoperative nausea and vomiting. IV: factors related to postoperative nausea and vomiting. Anesthesiology 1960: Cohen MM, Duncan PG, DeBoer DP, et al. The postoperative interview: assessing risk factors for nausea and vomiting. Anesth Analg 1994;78: Apfel CC, Rauch S, Goepfert C, et al. The impact of smoking on postoperative vomiting. Anesthesiology 1997;87: Hough M, Sweeney B. The influence of smoking on postoperative nausea and vomiting. Anaesthesia 1998;53: Chimbira W, Sweeney BP. The effect of smoking on postoperative nausea and vomiting. Anaesthesia 2000;55: Rowley MP, Brown TC. Postoperative vomiting in children. Anaesth Intensive Care 1982; 10: Sossai R, Johr M, Kistler W, et al. Postoperative vomiting in children. A persisting unsolved problem. Eur J Pediatr Surg 1993;3: Rose JB, Watcha W. Postoperative nausea and vomiting in paediatric patients. Br J Anaesth 1999;83: Song D, Greilich N, Tongier K, et al. Recovery profiles of outpatients undergoing unilateral inguinal herniorraphy: a comparison of three anesthetic techniques. Anesth Analg 1999;88:S Pusch F, Freitag H, Weinstabl C, et al. Single-injection paravertebral block compared to general anaesthesia in breast surgery. Acta Anaesthesiol Scand 1999;43: Eberhart LHJ, Seeling W, Staack AM, et al. Validation of a risk score for prediction of vomiting in the postoperative period. Anaesthesist 1999;48: Sneyd JR, Carr A, Byrom WD, et al. A meta-analysis of nausea and vomiting following maintenance of anaesthesia with propofol or inhalational agents. Eur J Anaesthesiol 1998;15: Borgeat A, Wilder-Smith OHG, Rifat K, et al. Adjuvant propofol is effective in preventing refractory chemotherapy associated nausea and vomiting. Anesthesiology 1992;3A: A344

18 30 Apfel and Roewer 33. Borgeat A, Wilder Smith OH, Saiah M, et al. Subhypnotic doses of propofol possess direct antiemetic properties. Anesth Analg 1992;74: Borgeat A. Antiemetic effect of propofol. Anaesthesia 1996;51: Scuderi PE, D Angelo R, Harris L, et al. Small-dose propofol by continuous infusion does not prevent postoperative vomiting in females undergoing outpatient laparoscopy. Anesth Analg 1997;84: Gan TJ, Glass PS, Howell ST, et al. Determination of plasma concentrations of propofol associated with 50% reduction in postoperative nausea. Anesthesiology 1997;87: Hartung J. Nitrous oxide it s enough to make you vomit [letter]. Anesthesiology 1993; 78: Hartung J. Twenty-four of twenty-seven studies show a greater incidence of emesis associated with nitrous oxide than with alternative anesthetics. Anesth Analg 1996;83: Divatia JV, Vaidya JS, Badwe RA, et al. Omission of nitrous oxide during anesthesia reduces the incidence of postoperative nausea and vomiting. A meta-analysis. Anesthesiology 1996;85: Thompson PI, Bingham S, Andrews PL, et al. Morphine 6-glucuronide: a metabolite of morphine with greater emetic potency than morphine in the ferret. Br J Pharmacol 1992;106: Costello DJ, Borison HL. Naloxone antagonizes narcotic self blockade of emesis in the cat. J Pharmacol Exp Ther 1997;203: Anderson BJ, Ralph CJ, Stewart AW, et al. The dose-effect relationship for morphine and vomiting after day-stay tonsillectomy in children. Anaesth Intensive Care 2000;28: Anderson BJ, Pearce S, McGann JE, et al. Investigations using logistic regression models on the effect of the LMA on morphine induced vomiting after tonsillectomy. Paediatr Anaesth 2000;10: White PF, Coe V, Shafer A, et al. Comparison of alfentanil with fentanyl for outpatient anesthesia. Anesthesiology 1986;64: Phitayakorn P, Melnick BM, Vicinie AF 3rd. Comparison of continuous sufentanil and fentanyl infusions for outpatient anaesthesia. Can J Anaesth 1987;34: Flacke JW, Bloor BC, Kripke BJ, et al. Comparison of morphine, meperidine, fentanyl, and sufentanil in balanced anesthesia: a double-blind study. Anesth Analg 1985;64: Jakobsson J, Davidson S, Andreen M, et al. Opioid supplementation to propofol anaesthesia for outpatient abortion: a comparison between alfentanil, fentanyl and placebo. Acta Anaesthesiol Scand 1991;35: Langevin S, Lessard MR, Trepanier CA, et al. Alfentanil causes less postoperative nausea and vomiting than equipotent doses of fentanyl or sufentanil in outpatients. Anesthesiology 1999;91: Apfel CC, Kranke P, Eberhart LHJ, et al. A comparison of predicting models for postoperative nausea and vomiting. Br J Anaesth 2002;88: Bonica JJ, Crepps W, Monk B, et al. Postanesthetic nausea, retching and vomiting. Anesthesiology 1958;19: Smessaert A, Schehr CA, Artusio JF. Nausea and vomiting in the immediate postanesthetic period. JAMA 1959;170: Gan TJ. Postoperative nausea and vomiting can it be eliminated? JAMA 2002;287: Junger A, Hartmann B, Benson M, et al. The use of an anesthesia information management system for prediction of antiemetic rescue treatment at the postanesthesia care unit. Anesth Analg 2001;92: Tramer MR, Fuchs-Buder T, Sansonetti A, et al. Low incidence of the oculocardiac

19 Risk Assessment of PONV 31 reflex and postoperative nausea and vomiting in adults undergoing strabismus surgery. Can J Anaesth 1997;44: Watcha MF, White PF. Postoperative nausea and vomiting. Its etiology, treatment, and prevention. Anesthesiology 1992;77: Rusch D, Happe W, Wulf H. [Postoperative nausea and vomiting following strabismus surgery in children. Inhalation anesthesia with sevoflurane-nitrous oxide in comparison with intravenous anesthesia with propofol-remifentanil]. Anaesthesist 1999;48; Saiah M, Borgeat A, Ruetsch YA, et al. Myopexy (Faden) results in more postoperative vomiting after strabismus surgery in children. Acta Anaesthesiol Scand 2001;45: Andrews PLR. Physiology of nausea and vomiting. Br J Anaesth 1992;69:2S 19S 59. Hovorka J, Korttila K, Erkola O. The experience of the person ventilating the lungs does influence postoperative nausea and vomiting. Acta Anaesthesiol Scand 1990;34: Hechler A, Naujoks F, Ataman K, et al. [The incidence of postoperative nausea and vomiting is not effected by routinely applied manual pre-oxygenation during induction of anesthesia]. Anasthesiol Intensivmed Notfallmed Schmerzther 1999;34: Hovorka J, Korttila K, Erkola O. Gastric aspiration at the end of anaesthesia does not decrease postoperative nausea and vomiting. Anaesth Intensive Care 1990;18: Trepanier CA, Isabel L. Perioperative gastric aspiration increases postoperative nausea and vomiting in outpatients. Can J Anaesth 1993;40: Cheatham ML, Chapman WC, Key SP, et al. A meta-analysis of selective versus routine nasogastric decompression after elective laparotomy. Ann Surg 1995;221: Pierre S, Benais H, Pouymayou J. Apfel s simplified score may favourably predict the risk of postoperative nausea and vomiting. Can J Anesth 2002;49: Kranke P, Apfell CC, Papenfuss T, et al. An increased body mass index is no risk factor for postoperative nausea and vomiting. A systematic review and results of original data. Acta Anaesthesiol Scand 2001;45: Haigh CG, Kaplan LA, Durham JM, et al. Nausea and vomiting after gynaecological surgery: a meta-analysis of factors affecting their incidence. Br J Anaesth 1993;71: Toner CC, Broomhead CJ, Littlejohn IH, et al. Prediction of postoperative nausea and vomiting using a logistic regression model. Br J Anaesth 1996;76: Apfel CC, Goepfert C, Sefrin P, et al. A risk score for the prediction of postoperative vomiting. Br J Anaesth 1997;78: Apfel CC, Rauch S, Goepfert C, et al. Predictors of postoperative vomiting: a model for risk assessment. Anesthesiology 1997;87:A Maleck WH, Piper SN, Apfel CC, et al. Predictive models for postoperative nausea and vomiting. Br J Anaesth 2002;89: Eberhart LH, Hogel J, Seeling W, et al. Evaluation of three risk scores to predict postoperative nausea and vomiting. Acta Anaesthesiol Scand 2000;44: Henzi I, Walder B, Tramer MR. Dexamethasone for the prevention of postoperative nausea and vomiting: a quantitative systematic review. Anesth Analg 2000;90: Subramaniam B, Madan R, Sadhasivam S, et al. Dexamethasone is a cost-effective alternative to ondansetron in preventing PONV after paediatric strabismus repair. Br J Anaesth 2001;86: Thomas R, Jones N. Prospective randomized, double-blind comparative study of dexamethasone, ondansetron, and ondansetron plus dexamethasone as prophylactic antiemetic therapy in patients undergoing day-case gynaecological surgery. Br J Anaesth 2001;87: Rajeeva V, Bhardwaj N, Batra YK, et al. Comparison of ondansetron and dexamethasone in prevention of PONV in diagnostic laparoscopy. Can J Anaesth 1999;46: Splinter WM, Rhine EJ. Low-dose ondansetron with dexamethasone most effectively